Magic Beans - by Venkatesh Rao - Ribbonfarm Studio

As you might know if you’ve been following me on Twitter, I’ve been heads-down exploring Web3 for the last month or so. Much of my initial experimentation has been focused on NFTs, or non-fungible tokens, which are sort of the entry drug of Web3. I’ve minted a handful of simple ones, both by myself and in collaboration with others. You can see the ones I’ve sold here. I’m starting to think about more complex experimental ones I might do, and have earmarked my earnings so far towards those experiments.

So what exactly do you buy when you buy an NFT? And on the selling side, what exactly do you sell?

Depending on who you ask, the answer is “nothing, it’s a scam!” and “a stake in the glorious Web3 future!”

And depending on the NFT in question, either answer, or even both answers, might be correct. That’s not surprising. Every technology has a scammy side and a high-potential aspect. Asking whether a technology is valuable or scammy is one of the oldest technology criticism scams.1

The funny thing is, critics who sneer, “why would you want to buy a JPEG” are unwittingly being too generous in their ignorance. The truth is you aren’t even buying the JPEG.

Hell, you aren’t even buying even an unstable pointer to a JPEG. Including something like a URL to the JPEG or an IPFS content identifier2 is simply a matter of artistic courtesy. Technically an NFT is simply a token representing an object. It is not required to point to it. The object need not even exist.

So is an NFT something like a scribbled IOU backed by no collateral and no enforceable contractual commitments?

Worse! It’s not even an IOU.

An NFT does not imply anything about rights assignation. The author of the work represented retains all rights. They may assign some rights when they mint the NFT (ie, put the rights within the immutable description on the blockchain), but that’s entirely up to them, and is not actually commonly done. In general rights are assigned after the fact, if at all.

In fact, paradoxically, there seems to be a perception of disreputable grifting attached to assigning rights upfront. (The paradox does not run deep. The reasoning is similar to why a guest post pitch for my blog with the line “it will be an original, CopyScape protected3 article” in it is 100% a spam blogger (splogger)).

Hilarious right? People buying these things must be crypto-rich idiots who aren’t doing their research and don’t realize how easy it is to copy digital objects, right?

Wrong.

By and large people in the NFT world are not just aware of these things, but often go around educating artists about all this, because artists new to being courted by technology are more likely to assume they’re actually selling their art. If this is a con, it’s one where the marks have to explain how to work the scheme to the would-be con artists, and convince them to run them.

So what is going on here?

There are three levels to mental models you’ll see tossed around about what NFTs are, conceptually. Answers to the question “what do you buy when you buy an NFT?”

The first level has to do with signification:

NFTs with more features, such as being an element in a large set of related ones (such as CryptoPunks, BoredApes, Loot or the latest one — Wolf Game) admit a second class of interpretations, such as:

These complex NFTs exploit some of the technical affordances of NFTs (which are contracts based on standards such as ERC-721 and the newer ERC-1155), to create richer behaviors. But the claims to value I’ve seen seem rather forced, and the explanations and justifications rather strained.

Related to these is a third class of mental models that rest on richness in future environments rather than in the NFTs themselves. Instead of building a complicated and expensive trading-card type game out of NFTs with complex programming attached, you do the regular selling of dumb not-even-JPEGs thing, but make the case that in some future environment, the infrastructure itself will be “NFT aware” and somehow assign you rights. For example:

That’s 3 classes of mental models for NFTs, which I summarize as follows:

Or in other words, the right to represent a thing, the right to do things with a thing, and the right to expect certain things in speculative environments. These mental models correspond to a scale of increasing value and increasing speculativeness.

The most fully realized example of all three mental models today is domain names with .eth extensions, managed on the blockchain by the Ethereum Name Service (ENS), an analogue to DNS. On Web3, .eth names function as NFTs (they are non-interchangeable tokens on the blockchain owned by specific parties). The details of how the three mental models apply are complex (ask your friendly neighborhood Web3 enthusiast to explain).

All these common mental models do capture aspects of what is going on. And they’re not mutually exclusive. Clearly there are ways you can mix and match these aspects.

But none of this explains the crazy valuations of NFTs, and more importantly, none of it is anything that must necessarily be done with NFTs. Each of the technical possibilities I’ve laid out could be realized in some other, cheaper way, such as with a database. ENS, for example, solves a problem already solved by DNS.

Yes, there is some elimination of trusted third parties, and depending on the setup, unique patterns of decentralization, openness, and market-like behavior. But again, for any specific example, you can probably think of a way to do it without NFTs.

“What problem does it solve” or even the slightly more imaginative “what problem could it solve” is another age-old technology criticism scam that goes nowhere fast.4

So let’s approach it from a different angle.

Over a decade ago, when Web2 was new, we all talked a lot about “social objects.” In 2007, an early social networking entrepreneur Jyri Engeström popularized the basic notion of social objects, Hugh MacLeod wrote a thing about them, and I did too, in 2009. For a while there was talk of “object-centered sociality” and in consulting gigs I talked of “high-frequency social objects” (ie things like YouTube videos and early memes).

NFTs clearly belong in that tradition. They are digital social objects. Except they are a lot wilder. They don’t live within particular platforms or walled gardens, and aren’t attached to a particular company’s technology or even a specific open standard.

They are technically complex, but socially promiscuous in a way only much simpler digital things, such as plain text files or JPEGs, have been so far. As we’ve discussed, they’re also kinda content-free. They bear no necessary payloads, pointers, programmed affordances, or rights. All those things are optional extras. The NFT itself is just a token whose only necessary property is that it be uniquely identifiable, ownable, and tradeable. It is something like a digital-object zero (which explains both the moral panic and the excitement).

So the parsimonious mental model of an NFT based on what we’ve discussed so far is that it is a promiscuous and vacuous digital social object. A thing that contains no substance itself, not even reliable pointing properties, but has an identity and can be traded, transferred, mixed and matched indiscriminately with other things in a social context.

To explore what an NFT is, I tried to imagine what the opposite of an NFT might be, while still remaining a digital object. An anti-NFT. If an NFT is socially promiscuous and vacuous, an anti-NFT ought to be a non-promiscuous and substantial digital social object.

To complete the 2x2, a promiscuous and substantial digital social object would be a valuable work in the public domain. A non-promiscuous and vacuous object might be something like a vague desire to write a novel. There’s no substance to it until you actually start making up a story and begin the writing, and it’s not transferrable or otherwise available for use as a social object. At best you might mumble at a party, “I’m working on a novel.”

In this 2x2, the most mysterious object is actually the anti-NFT, not the NFT. Let’s try and imagine one with a thought experiment. A bit of legal science fiction concerning something that’s approximately an anti-NFT.

A manuscript for a previously unpublished science fiction novel by Isaac Asimov has just been discovered. Scholars who have seen and authenticated it claim it is his best work. But no rights to it have been exercised as yet. It is a thing that has known, clear value, via reference to previous bestselling works by Asimov, but the value only exists in potentia. It is pure possibility attached to clear value.

But for some weird reason, the Asimov estate offers you all rights to it for $1000, except for one big class of rights — you cannot transfer, assign, rent, resell, or share the rights. You can only exercise them yourself.

Unlike a typical NFT, there is something real, and of clear value here. It is an actual asset instead of a token representing one. And unlike a typical NFT, it cannot be bought or sold, only held and exercised. Here you have socially inert rights to a thing of substance, whereas with NFTs you have a socially promiscuous rights to a thing of apparently no substance.

Would you want it? Or is it like a white elephant gifted you by the King of Thailand — something precious and holy you’re not allowed to actually use in any way, and might bankrupt you to maintain (this is where the term white elephant comes from)?

This is a trickier question than it might seem. I’m not a lawyer, let alone one versed in copyright law, but let me take a stab at working out the implications of this premise.

If you cannot assign the rights to anyone else, you cannot use the rights as collateral to raise capital. So for example, if you wanted to raise financing to publish the book, or make a movie based on it, your investors would have no recourse if you screwed up the project. They couldn’t claim your rights and have someone else publish the book or make the movie. Let’s assume the courts are also powerless somehow to take away the rights.

Studios couldn’t option the movie rights because you can’t actually assign them. If you decided to fund the movie yourself, and wanted to hire writers and artists to work on it, you could only do so under pure work-for-hire arrangements where you own their work outright. More generally, you’d be forced to outright buy anything that used the IP you own, because it can’t co-mingle with other IP easily. You couldn’t take on equity partnerships because that would mean sharing rights to derivative works with partners. You couldn’t create extended-universe crossover connections with published Asimov works like the Robot or Foundation stories.

These are not purely theoretical concerns. Versions of this sort of thing already play out. For example, it took over a decade for Spider-Man to show up in the Marvel Cinematic Universe because the rights were with Sony, and brokering a deal proved to be really hard. Our thought experiment is an extreme version of this sort of situation.

With a team of clever lawyers, you could probably still figure out a way to raise money based on your rights and do something with them, but it would be a serious pain and considerably increase the risks. You’d probably have to do a lot more yourself — handle self-publishing, write bad scripts, do bad direction, and so on.

The deal would probably still be worth it in this case. $1000 buys you at least one major right that requires no skill to exercise — the right to block others from doing anything, and that’s probably worth a lot by itself. If the manuscript leaks, you could go around suing people who tried to publish bootleg versions.

The point of this legal science-fiction thought experiment is that copyright works in the somewhat promiscuous way it actually does because it’s a pragmatic way to enable creative collaboration. Every creative industry has various conventional mechanisms and practices allowing rights to works to be bought and sold, assigned, shared, used as security for financing and so on. Rights that are stripped of this social-legal character are much less valuable.

Creative works are social objects in the fullest sense — they exist in a nexus of social, cultural, and legal links to other people and things, and they are only valuable to the extent they can fully participate in their whole context. Rights regimes restrict known rights, but also create rights if they are well designed, driving generativity.

So in general, rights are useless if you do not have the skill to do anything with them, which is why there is a need to be able to trade with them. And the more subtly you can specify and trade rights, the more you can do with them.

An anti-NFT would have the approximate opposite market behavior of an NFT — clear value, but no way to make money off it.

Or to put it in less financial terms, a creative work is a creature of context, and it can’t really achieve a full expression of itself unless it can fully interact with its context.

The job of an NFT is to create and embody this context. To understand how this might work, let’s first look at how regular rights work.

A classic joke illustrates the subtleties of ownership and rights in a social context.

A rube from a village goes to the big city and sees a skyscraper for the first time. He is impressed and begins counting the floors under his breath. A city grifter comes up to him pretending to an official, and declares that there is a fee for counting floors, $10 per floor, and demands to know how many floors the rube has counted, and to be paid accordingly. The rube forks over $100 for counting 10 floors.

The grifter walks away, pleased with his score.

The rube later brags to his friends back at the village: “Hah, I pulled a fast one there, I’d actually counted to 20 floors!”

The joke gets at why the so-called “right clicker” problem — basically just right-clicking to save a copy of an NFTed artifact for yourself — is only funny if you genuinely don’t know what you’re talking about and think people buy NFTs to exclude others from all rights to the represented thing.

Right-clicking is not a problem for the NFT world because securing exclusive possession rights is not the problem NFTs aim to solve.

Right-clicking and saving a JPEG is about as valuable a piece of the rights-pie associated with a JPEG as counting the floors of a building. Being able to live in the building is a far more valuable right. Being able to make a movie based on a character depicted in a JPEG is a far more valuable right.

The reason people think they’re being clever is because they are applying not-even-wrong mental models where physical possession is 9/10ths of the law, a principle that rests on the more basic fact that for dumb physical objects, physical possession unlocks 9/10s of the value. So if you can defend possession (either through the law, or with a gun), you defend the value.

Here is a principle:

The smarter an artifact, the smaller the fraction of rights associated and defensible with, and unlocked by, physical possession.

NFTs don’t assign any digital rights because rights as they currently work are too much of a blunt instrument, thanks to be rooted in notions of physical possession. To the point where asserting and assigning some rights legibly might actually destroy less legible future rights. NFTs aim to create a much richer calculus of digital rights — one centered on the sociality of objects rather than physical possession.

DRM as traditionally construed is a naively skeumorphic attempt to solve for physical possession of a digital artifact. Not only does this not make sense for purely digital objects, it does not even make sense for physical objects.

This is easy to see in the case of physical objects with a significant software component — owning a SaaS appliance device physically does not mean you control it physically. Over-the-air software updates, remote administration capabilities, and the need for specialized skills to exercise physical agency mean possession is not 9/10ths of either the value or the law. This is true to such an extent that businesses offering SaaSy appliance products often give away the physical objects below cost or even for free. All their money comes from subscription services based on rights they retain.

This even extends to things not eaten by software. If you own a combination lock but don’t know the combination, do you own it? Does someone else who does not possess it physically, but knows the combination, in fact have greater agency over it?

Still too crypto for you? Combinations are after all just physically embodied cryptography, right?

How about a donut? Dumb enough for you?

You have a donut. You’re the only one who can eat it. Congrats. Did you get 100% of the value?

Do you have the donut recipe? Can you reverse-engineer it? Can you run a chain restaurant selling those donuts? Can you build a business empire to compete with Dunkin’ Donuts based on currently possessing a single sample of their donuts? Can you put up a storefront with the Dunkin’ Donuts logo on it? Negotiate with flour and sugar suppliers with the bargaining clout of Dunkin’?

Possession is 9/10ths of the value of a thing only in extraordinarily rare situations involving extraordinarily dumb objects. Even unique art objects don’t really obey this principle.

Do you really think 9/10ths of the value associated with the Mona Lisa accrues to the Louvre? I can put a photograph of it in an article, write a book discussing it, even make a movie about it, without paying the Louvre anything. The Louvre has the right to hold the original physically, but every other meaningful right is in the public domain. Not only are these other rights collectively far more valuable, physical possession only has value by way of derivation from these other rights. Obscure paintings nobody else cares about are generally not worth possessing. That’s why there is no market for crayon drawings by kids stolen from refrigerator doors.

Another example. Let’s say you download a DRM-free bootleg copy of Asimov’s Foundation. Yay, you saved yourself a few bucks and put one over those evil copyright owners who want you to pay money for bits.

What have you actually acquired by way of possession of a physical copy? How much of the value of the book — not just your copy of it, but the book itself, as a social object that has already induced a vast social context around it over decades — have you unlocked by taking possession of a single physical copy (digital or paper)?

Can you make a Foundation movie? Can you publish your own edition and sell a million copies of it? Can you edit the book to expand it with your own fan-fiction inserts and sell it? Can you run a successful Asimov fan event?

Those things would be far harder to do without getting the associated rights. The $10 you saved by acquiring a bootleg copy is barely a rounding error relative to the value of the rest of the rights in play. At best, you acquired the right to read the book and talk about it without paying $10. And if you talk about it enough, the publisher will make back more than $10 in legal sales.

Making NFTs “do” physical-possession DRM would not actually be that hard of a problem to solve if that were actually a problem worth solving.

To build on the skyscraper joke: Imposing and collecting a fee for counting the floors of a skyscraper would also not be that hard of a problem to solve if it were actually worth solving.

You could set up NFTs such that only those who hold the token can decrypt the file it represents, and set things up so that piracy is kept to manageable level.

This is probably going to be one of the use cases of NFTs (one I’m exploring myself — using an NFT as a way to do a paywall). But it’s almost certainly going to be a minor use case.

The far more exciting potential of NFTs lies in managing rights that are not a function of possession, but of social context.

NFTs aim to ground a much larger set of rights to artifacts in something less flimsy than possession.

That something is latent in the social context that might grow around any creative artifact. Every object is a social context in potentia, and NFTs represent unspecified rights in that context. That context may or may not emerge, so an NFT is also a bet that it will. And like buying stocks, buying NFTs increases the likelihood of the context actually emerging.

Setting aside the question of whether or not DRM is the right thing to do to digital artifacts, how do you do DRM right, in a way that doesn’t just respect the promiscuously social nature of creative works, but actually unlocks and amplifies then in meaningful ways? In a way merely controlling possession of copies cannot? In a way that even just releasing a work to the public domain might not?

We saw hints of the answer with things like crowdfunding, where creators set funding goals and promised rewards at different levels. But that’s still too rooted in possession-based thinking.

Old DRM is an ineffective blunt instrument, attempting to manage all rights indirectly by policing patterns of possession — the most legible right. In our skyscraper joke from the previous section, this is a bit like trying to manage occupancy rights to a building by policing the right to count the number of floors it has.

Possession-based approaches often secure low-value legible rights at the expense of destroying swathes of high-value illegible rights.

And the smarter the artifact, the truer this is. Which is why rights are often not enforced in practice. Big-name authors do not generally clamp down on fan fiction for example.

I hope the discussion so far has made the case that the point of NFTs is managing social rather than possession-based rights. Mere possession, you might say, is the freemium right. A right you give away for free in most cases, or only police and meter with a minimal level of effort, and only so long as it doesn’t get in the way of exercising more complex rights.

Analogy: a real-estate developer might put up a paid parking lot on a plot of land, but then make parking free on the same plot if they proceed to build a multi-level retail shopping space on that same lot. Unlocking a higher kind of value might make a lower kind of value not worth policing.

How does this work with non-possession-based rights for digital social objects?

The brief history of NFTs suggests that there are three kinds of emerging rights that can potentially go with any digital social object:

To manage any of these rights, a notion of ownership, based on scarcity, is necessary. Critics who think think actual scarcity is being created out of abundance miss the point entirely. The scarcity is on a map, not on a territory. In other words, the rights have value because they are defined relative to a social context, and might only exist in that context. NFTs are something like relativistic objects. What they are worth is a function of what the reference social-space-time coordinate system is.

This context is what creates the scarcity.

A pre-crypto example is domain names. Nothing stops you from naming your website “cnn.com” and putting up a banner declaring that name. The problem is, a particular “map” — the view of the internet maintained by the domain name service (DNS) — is set up to only recognize one referent for a name. The right to a domain name is really a right to a map reference in the social context created by DNS. The reason it is valuable enough to pay for is that it is the only map available to navigate the internet. You could make up your own map with different names for all websites. If you can get people to use it, you have a valuable social context of scarcity.

A scarcity-creating social context is a bit like a language. You can make up your own language. If you can get people to speak it, you have a valuable social context of scarcity. English texts have value because that’s a very valuable social context. Esperanto and Klingon texts have less value because those languages are spoken by very few people.

NFTs take this sort of logic to another level. Instead of a few such global scarcity-creating contexts, any object can create its own scarcity-based social context. Every object can create a map of its own social future, when it represents nothing more than virgin territory, and can sustain nothing smarter than a land-grab race driven only by vague indications of the landscape available for occupation.

You could say “every object can be a social network” except that social networks are a more primitive Web2 concept. The social contexts induced by NFTs are already much richer than mere social networks. Instead of supporting a few narrow social-networking protocols, they allow you to program any kind of behavior you want into the social context.

Social networks are a special case of social context. NFTs operationalize control of the social context of artifacts in general ways. Just as the internet is a “network of networks,” a blockchains is a “context of contexts.”

“Ownership” on the blockchain operationalizes control of digital objects in material and social ways. This is a richer concept than physical ownership.

Why?

Consider how the social context of ownership is managed in the physical world.

A key to a lockable door operationalizes control of a house in terms of physical access, but a title or deed operationalizes control in a different way — on the “map” of ownership used by banks to decide who has the right to use the house as collateral for a loan.

The story doesn’t end there. Neighborhood associations might operationalize control in other ways, by policing building activities to prevent obstruction of views for example. If a stream flows through your property, a water-rights-management regime might operationalize control of rights to water by policing pumping, diversion, and damming activities.

As more becomes known about a material reality, more rights must be handled. Airspace rights, radio spectrum rights, pollution rights…

Societies manage this on a cross-the-bridge-when-we-come-to-it basis. Just because you own a piece of land today doesn’t mean you’ll own the rights to wormhole portals that open up on it tomorrow.

This isn’t a bad way, but is rather reactive.

NFTs proactively create future social contexts on a speculative basis. The Loot NFT project, for example, has created a partial social context for a future role-playing game. The value of the NFT is a function of the quality of the speculation it embodies.

The traditional approach is also bad in a different way — it privileges currently powerful parties. When a new right is finally legible enough to bet on, it is typically too valuable for anyone but the already wealth to bet on. By institutionalizing rights at an earlier, more nebulous stage, NFTs allow more people to participate. NFTs allow over-the-horizon rights to be bought when they are very cheap, by those with the instincts to bet on them.

Not convinced? The first round of buyers for one of the most storied NFT collections around, CryptoPunks, actually got them for free. They simply had to get in the game early enough, and pay transaction fees.

An NFT is something like a combination of a title or deed and a key. It doesn’t just represent your right to a digital object, but materially and socially operationalizes control of it across all sorts of undefined and arbitrary potential social contexts. It secures a place for you in a future possible world. And the earlier you see the possibilities, the cheaper it is to buy in.

Most worlds will fail to actually emerge of course. Most startups fail. Most new restaurants fail. Most novels fail.

Most NFTs will turn out to be worthless.

But some small fraction of the time, NFTs will serve as seeds for entire universes of possibility. The risk of the NFT game, like the risk of the VC game, is not that you’ll lose money on the duds, but that you’ll miss out on the non-duds.

What’s a good mental model for this?

Let me now justify the title of this essay.

Appropriately for a technology based on something called “Ethereum,” NFTs provide a richer means to suggest the contours of an attractive future, thereby raising the resources required to actually create it.

NFTs are a better kind of vaporware, and that’s a good thing.

Unlike dubious architecture diagrams for unwritten software, grant proposals, mockups, or even the famed whitepapers of the the ICO boom of 2017, NFTs paint an impressionistic portrait of what might be, using what are effectively design fictions. And through the valuation of pieces of that portrait, demonstrate social proof of future value.

That’s a better kind of vaporware.

But vaporware is a weak metaphor. We need a better one. I propose magic beans.

An NFT represents an access pass to an unspecified, generative possible future associated with an object. It is a key to a possible world. A key that is distinguishable from other keys. A key to a world that may or may not open up. A key that carries some hints of the kinds of locks it might unlock. A key that, unlike fungible tokens (or stock), allows you to place educated bets on different bits of the future being gestured at.

It is a magic bean, as in the fable of Jack and the Beanstalk.

If you recall, in that story, Jack is delegated by his poor family to go to the market and sell their one cow for cash, but Jack, being impressionable and gullible, falls for the sales pitch of a mysterious stranger, and returns with a handful of supposedly magic beans. His father is outraged, and yells at Jack and throws the beans out the window.

The next morning, they wake up to find that a giant beanstalk has grown where the magic beans fell, reaching into the clouds. Jack climbs the beanstalk to find the realm of a giant. He steals a bunch of treasure and climbs down, and when the angry giant tries to follow him down, he chops down the beanstalk. The giant comes crashing down and is killed.

Depending on the version of the story, the giant may or may not have deserved it.

This is, I think, a rich allegory for the NFT hustle. It accommodates interpretations of real future value (treasure!) as well as real future scams (Jack is a thief! He steals! He’s possibly a murderer of an innocent giant!). It suggests much richer, more expressive ways of investing in future possibilities than through traditional investment vehicles like VC funds or Kickstarters, and doing so earlier, with fewer resources. Possibly no resources.

It suggests a marketplace of ideas where most beans sold as magic beans are not in fact magical, but some are. It even suggests how we ought to think of the big FAANG platforms — as the bad giants whose realms we ought to be stealing treasure from.

Will any of this work out? I don’t know. But it’s interesting enough to bet on.

The scam rests on the pretense that technological progress is “human-centered” and humans choose to create and adopt only “good” technologies. Depending on who is doing the pretending, this is either a scam, sloppy is-ought confusion, or childish conceit.

A native storage layer abstraction for Web3, where content is identified directly, based on intrinsic properties (hashes) rather than via a location address. This allows for a slightly less fragile way to reference content, but is still dependent on someone (ideally many someones) bothering to keep the content alive on the network.

A plagiarism detection service that makes sense to use if you’re dealing with large volumes of content with any degree of automation, but hardly a meaningful feature for a blog guest post pitch.

This scam rests on the pretense that technology evolves as a series of solutions to problems. It does not. It would be slightly more accurate to say that technology evolves by creating new problems worth solving.